Our goal is to identify and quantify specific neuroimaging markers of Alzheimer Disease (AD) using an automated brain MRI reading technique, the automated labeling pathway (ALP). ALP is a reliable, easy to operate, fully automated technique that generates regional and global volumetric measures of the brain. This self-contained project proposes secondary analysis of 532 brain MRIs acquired in Pittsburgh in 1997-99 as part of the Cardiovascular Health Study (CHS; N01HC85082), a multisite observational study to investigate cardiovascular risk factors in the elderly. CHS participants received a detailed dementia evaluation as part of the Cognition Study of the CHS (CHS-CS; R01AGO15928). Our project will take advantage of the wealth of epidemiologic data collected from 1989 to date, including the follow-up dementia evaluation conducted in Pittsburgh by Dr. Lopez (R01AG020098). ALP volumetric measures of the brain will enrich the CHS-CS dataset and will be available to other CHS investigators. We propose that ALP- measures of global and focal atrophy (hippocampal and medial temporal lobe): 1) differentiate older individuals who are cognitively normal, vs. those who are affected by AD or Mild Cognitive Impairment; 2) discriminate cases of dementia better than conventional CHS-CS brain MRI measures (ventricular and white matter grade); 3) discriminate those cognitively normal adults who convert to dementia over the following 5 years vs. those who convert to MCI or remain cognitively normal. The strengths of this application include: immediate availability of 532 brain MRIs in Pittsburgh, well documented cases of dementia from 1989 to- date, local support for the use of ALP, extensive demographic, genetic and clinical data. In accordance to the Final Consensus of the NIA Alzheimer's initiative, we will evaluate the convergent validity of measures from novel and conventional brain MRI reading methods, and will test if diagnosis of dementia can improve when specific neuroimaging biomarkers are included in the diagnostic criteria. We also plan to optimize and support the use of the ALP method by other investigators. The results of this proposal will lay a foundation for a future RO1 application to assess neuroimaging markers of dementia risk in a sample of over 3,000 CHS participants, for whom repeated MRIs and over 15 years follow up data on dementia are available. [unreadable] [unreadable] [unreadable]